Influence of cyanobacteria, mixotrophic flagellates, and virioplankton size fraction on transcription of microcystin synthesis genes in the toxic cyanobacterium Microcystis aeruginosa

Abstract Toxic cyanobacteria such as Microcystis aeruginosa are a worldwide concern in freshwater reservoirs. Problems associated with their mass occurrence are predicted to increase in the future due to global warming. The hepatotoxic secondary metabolite microcystin is of particular concern in this context. This study aimed to determine whether co‐occurring microorganisms influence the expression of microcystin biosynthesis genes. To this end, we performed cocultivation experiments and measured mcyB and mcyD transcripts in M. aeruginosa using RT‐qPCR. We utilized representatives from three different plankton groups: the picocyanobacterium Synechococcus elongatus, the unicellular flagellate grazer Ochromonas danica, and virioplankton from two different lakes. The presence of S. elongatus significantly increased mcyB and mcyD transcription in M. aeruginosa. Cocultivation with the mixotrophic chrysophyte O. danica did not increase the transcription of mcyB and mcyD; in fact, mcyD transcripts decreased significantly. The virioplankton size fraction of environmental water samples induced a significant increase in mcyB and mcyD transcription when obtained from lakes with cyanobacterial blooms. Our results show that co‐occurring microorganisms influence the expression of microcystin biosynthesis genes in M. aeruginosa.

Cyanobacteria serve as prey for unicellular zooplankton. For instance, protists such as the mixotrophic flagellate Ochromonas sp.

evaluated
Ochromonas as a tool to control Microcystis blooms.
Besides being grazed on by protists, lysis by bacteriophages is a major mortality factor for cyanobacteria (Proctor & Fuhrman, 1990).
The biological function of microcystin is still under investigation and while its role in the survival of the Microcystis cell is not yet fully understood, there have been several foci of research over the years: Microcystin, a peptide toxic to many animals, was suspected to act as a defense against grazers such as Daphnia (Kurmayer & Jüttner, 1999;Pineda-Mendoza et al., 2014). Other studies, however, provide phylogenetic evidence (Rantala et al., 2004) as well as proteomic and physiological data (Zilliges et al., 2011) that suggest that a defense against metazoan grazers is not microcystin's primary function. Nevertheless, those grazers might have an effect on the expression of microcystin biosynthesis genes (Pineda-Mendoza et al., 2014). In addition, some doubt has been cast on microcystin as a primary defense mechanism against unicellular grazers such as mixotrophic flagellates (Wilken et al., 2010), albeit the situation is less clear here. Instead, more recent studies have focused increasingly on the intracellular and regulatory functions of microcystin within Microcystis cells (Makower et al., 2015;Meissner, Fastner, & Dittmann, 2013;Zilliges et al., 2011).
However, the discoveries that microcystin has multiple intracellular effects and that defense against metazoan grazers is not its primary role do not rule out the possibility that microcystin regulation might be affected by secondary metabolites of other microbial community members (viruses, protists, or other cyanobacteria). This is suggested by observations that toxins may increase in the presence of other microorganisms (Vardi et al., 2002). For this reason, it is of the utmost importance to understand how interaction between members of the microsphere and toxigenic cyanobacteria affects their expression of toxicity genes. In addition, any biological agent considered for use in the control of harmful algal blooms should first be scrutinized for its potential effects on toxin gene expression to prevent exacerbation of the problem.

| Sampling and preparation of environmental samples
Water samples for experiments with the virioplankton size fraction were collected from Lake Klostersee and Lake Bergknappweiher, two Bavarian lakes frequently presenting with cyanobacterial blooms. Lake Klostersee (GPS coordinates 48.08, 11.96) is a polytrophic artificial lake with a maximum water depth of 2.5 m, located 545 m above sea level in Ebersberg, Germany (Gesundheitsamt Landratsamt Ebersberg, 2014). Lake Bergknappweiher (GPS coordinates 47.85, 11.23) is a small meso-eutrophic lake with a maximum water depth of 2.5 m, located 617 m above sea level, and about 50 km south-west of Munich, Germany (Teubner et al., 2004). Water samples (4 L) were collected on 25 June 2015 from Lake Klostersee and on 27 July 2015 and 27 August 2015 from Lake Bergknappweiher. Samples were taken from surface water near the shore, and bacterioplankton and zooplankton were removed by filtering the water through a 0.2μm pore-size cellulose nitrate filter (Sartorius, Göttingen, Germany). The filtrate containing the virioplankton and other particles smaller than 0.2 μm as well as dissolved organic matter is referred to as 0.2 μm filtrate in this study.
Subsequently, cross-flow filtration was performed to concentrate the particles present in the 0.2 μm filtrate (Vivaflow50, 10 000 MWCO, Sartorius, Göttingen, Germany). We obtained 100 ml cross-flow filtrate from each liter of 0.2 μm filtrate. This 10-fold concentration of particles larger than 10 kDa is referred to as particle concentrate. The particle concentrate contains all substances present in the 0.2 μm filtrate and is enriched in particles such as viral particles. For some experiments 0.2 μm filtrate was inactivated by means of autoclaving (121°C, 20 min) and is referred to as autoclaved filtrate.

| Cultivation and strains
The microcystin-producing cyanobacterium M. aeruginosa SAG14.85 and the mixotrophic flagellate Ochromonas danica SAG933-7 were obtained from the Culture Collection of Algae at Göttingen University in Germany (SAG). Ochromonas was selected for being a phytoplankton grazer. The picocyanobacterium S. elongatus PCC7942 was obtained from the Pasteur Culture Collection of Cyanobacteria at the Institut Pasteur in Paris, France (PCC). Synechococcus was selected due to its almost ubiquitous distribution and its ability to compete for similar resources as Microcystis (Stockner et al., 2002). Microcystis aeruginosa and S. elongatus cells were grown separately for 10 ± 2 days in 100 ml volumes of BG-11 medium (Rippka, Deruelles, Waterbury, Herdman, & Stanier, 1979) supplemented with 0.5 mmol/L ammonium chloride in 300 ml borosilicate Erlenmeyer flasks. Both strains grew unicellular and were precultured under the same experimental conditions. Optical density at 730 nm (OD 730 ) was measured with a spectrophotometer (model 150-20, Hitachi, Chiyoda, Japan) to monitor cyanobacterial growth. Growth rates for M. aeruginosa and S. elongatus without treatment were 0.15/day and 0.13/ day, respectively. Spent medium from S. elongatus cell cultures was obtained by filtering cultures through a 0.2μm pore-size cellulose nitrate filter (Sartorius, Göttingen, Germany) to remove any cells from the medium.
Ochromonas danica cells were grown in 50 ml Ochromonas medium (Röderer, 1986) in 100 ml borosilicate Erlenmeyer flasks. Ochromonas danica was grown for 5 days to a concentration of 1.5 ± 0.5 × 10 6 cells/ ml, and its growth was monitored by counting cells immobilized with  Table S2.

| Plaque assay
Plaque assay was performed as described previously (Millard, 2009) with modifications to accommodate bacteriophages from a freshwater environment. In short, concentrated log phase M. aeruginosa cells were mixed with 0.2 μm filtrate and grown on 1% (w/v) agar plates with 0.4% (w/v) top agar. Plates were prepared with washed agar agar (Carl Roth, Karlsruhe, Germany) and BG-11 medium (Rippka et al., 1979) supplemented with 0.5 mmol/L ammonium chloride.

| Humic acid treatment
To mimic the effect of the humic acids present in Lake

| Harvesting and RNA extraction
Harvesting of cell cultures and RNA extraction were carried out as described previously (Scherer, Raeder, Geist, & Zwirglmaier, 2017).
Briefly, 20 ml liquid culture was filtered through a 0.2μm pore-size cellulose nitrate filter (Sartorius, Göttingen, Germany). RNA from the filter was extracted as described in Penn, Wang, Fernando, and Thompson (2014), cleaned using the RNA Clean & Concentrator kit (Zymo Research, Irvine, USA), and quantified and quality checked using a NanoVue Plus spectrophotometer (GE healthcare, Little Chalfont, UK) and 1% agarose gel stained with GelRed (Biotium, Hayward, CA, USA), respectively.

| RT-qPCR
Relative quantification of transcripts by reverse-transcriptase quantitative polymerase chain reaction (RT-qPCR) and subsequent data analyses were performed as described previously (Scherer et al., 2017). Two target genes, mcyB and mcyD, representatives of the microcystin synthetase gene cluster, and four reference genes (rpoC1, gltA, rpoD, and GAPDH) were used in this assay. Suitability of the four reference genes as reference gene panel for each experimental condition was confirmed with geNorm (M < 0.5). For primers and cycling conditions refer to Scherer et al. (2017). Normalized expression was determined as described in Vandesompele et al. (2002). In order to determine relative normalized expression, expression level of the control samples is set to one.

| Statistical analysis
To test for statistically significant changes in gene expression, foldchange values were tested for normal distribution (Shapiro-Wilk test).
Subsequently, the data were analyzed using one-way ANOVA with the level of significance defined as p ≤ .05. If applicable, Tukey's post hoc test was performed. All statistical analyses were performed using PAST v3.10 software (Hammer, Harper, & Ryan, 2001).  Phelan and Downing (2014) and Hu, Liu, and Li (2004). In our study, however, growth curves failed to reveal any effect of spent medium from M. aeruginosa cultures on the growth of S. elongatus cultures (data not shown). (p = .049) but not for 10 3 cells/ml O. danica (p = .078). Transcription of mcyB was also decreased, but the difference was not significant (Table   S1). These results show that the eukaryotic grazer O. danica does not induce the upregulation of microcystin biosynthesis genes but does even cause downregulation of relevant genes at the concentrations tested here. Therefore, the hypothesis that microcystin constitutes an inducible defense mechanism against unicellular grazers was not confirmed. Our results rather support the findings of others who found F I G U R E 1 Relative normalized mcyB (left) and mcyD (right) mRNA expression in Microcystis aeruginosa. Error bars show standard error of the mean. *Statistically significant difference from control treatment, p ≤ .05. Cocultivation of M. aeruginosa with cell-free spent medium from Synechococcus elongatus culture (gray) or S. elongatus cells in spent medium (white) for 48 hr. The control treatment, BG-11 medium, was used for normalization no evidence that microcystin acts as a defense against mixotrophic flagellates (Wilken et al., 2010). The possible role of microcystin as an infochemical has been discussed previously (Schatz et al., 2007), and using microcystin as an intercellular signal in the event of grazing pressure could potentially benefit the M. aeruginosa population. In light of this discussion, our findings are all the more surprising. However, we cannot exclude the possibility that another molecule serves such a signaling function.

| Response to mixotrophic flagellates
In our study, O. danica did react to the presence of M. aeruginosa with a behavioral change like the one described by Pfandl, Posch, and Boenigk (2004). When no prey M. aeruginosa was present, the chrysophyte remained highly motile, moving around rapidly and frequently (Video S1). In the presence of M. aeruginosa cells, however, O. danica was mostly attached to the substratum and created a current with its flagella to capture prey bacteria (Video S2). The pivotal assumption that O. danica SAG933-7 does prey on M. aeruginosa SAG14.85 was verified by microscopic observation of engulfed M. aeruginosa cells ( Figure S1), which is in agreement with previous studies (Van Donk et al., 2009;Wilken et al., 2010).
Flagellates are thought to be responsible for more bacterial mortality than phages (Bettarel et al., 2003). This, and the fact that we did not observe an increase in toxin biosynthesis gene transcription in the presence of O. danica, suggests that the use of hetero-or mixotrophic flagellates is an interesting avenue to explore in the pursuit of controlling toxic cyanobacterial blooms.

| Effects of the virioplankton size fraction from Lake Bergknappweiher before and during algal bloom
Cultivation with 0.2 μm filtrate or particle concentrate from nonblooming Lake Bergknappweiher did not increase the transcription of mcyB or mcyD significantly (Figure 3a, Table S1). However, 0.2 μm filtrate and particle concentrate from Lake Bergknappweiher sampled during an algal bloom did cause significant upregulation of microcystin biosynthesis genes in M. aeruginosa ( Figure 3b). The cyanobacterial bloom was characterized by macroscopically visible phytoplankton aggregates on the water surface and a mass development of Microcystis-like and Dolichospermum-like cells.
The mcyB mean fold-change values of M. aeruginosa cultures exposed to 0.2 μm filtrate (p = .003) or particle concentrate (p = 0.002) differed significantly from that of controls. However, there was no significant change in cultures exposed to autoclaved filtrate (p = 0.240). Similarly, a statistically significant increase in mcyD transcripts of M. aeruginosa cultures exposed to 0.2 μm filtrate (p < .001) or particle concentrate (p < .001) was observed. For mcyD, even autoclaved filtrate caused F I G U R E 2 Relative normalized mcyB (left) and mcyD (right) mRNA expression in Microcystis aeruginosa. Error bars show standard error of the mean. *Statistically significant difference from control treatment, p ≤ .05. Microcystis aeruginosa cocultivated with 10 3 cells/ ml (gray) or 10 4 cells/ml (white) Ochromonas danica for 48 hr. The control treatment, Ochromonas medium, was used for normalization. Ochromonas danica concentrations chosen were 10 3 or 10 4 cells/ ml, as used in previous studies (Guo & Song, 2010;Van Donk et al., 2009) F I G U R E 3 Relative normalized mcyB (left) and mcyD (right) mRNA expression in Microcystis aeruginosa. Error bars show standard error of the mean. *Statistically significant difference from control treatment, p ≤ .05. The control treatment, BG-11 medium, was used for normalization. (a + b) Cocultivation of M. aeruginosa with autoclaved filtrate (black), 0.2 μm filtrate (gray), and particle concentrate (white) from Lake Bergknappweiher sampled before (a) and during (b) a cyanobacteria bloom for 48 hr. (c) Cocultivation of M. aeruginosa with 0.2 μm filtrate (gray) and particle concentrate (white) from Lake Klostersee sampled during a cyanobacteria bloom for 48 hr The presence of humic acid in Lake Bergknappweiher water was taken into account in a separate experiment. The change in water color due to the presence of humic acid, at two different concentrations, did not alter the transcription of microcystin biosynthesis genes significantly ( Figure S2 and S3). The results show that the humic acid preparation tested in this study does not influence the transcription of selected toxicity genes and suggest that the observed changes in gene transcription after treatment with 0.2 μm filtrate and particle concentrate are not caused by humic acid in the water. Whether the chemical composition of the humic acid in Lake Bergknappweiher water influenced results, is less clear and needs to be subject of further study.

| Effects of the virioplankton size fraction from Lake Klostersee
Particle concentrate but not 0.2 μm filtrate from Lake Klostersee fold-change values in cultures exposed to particle concentrate differed significantly from those exposed to 0.2 μm filtrate (p = .049) but not control cultures (p = .064). These results show that microcystin synthetase gene transcription in M. aeruginosa increased when the cyanobacterium came into contact with small molecules and concentrated particles from water taken during an algal bloom in Lake Klostersee.

| Effective components in the virioplankton size fraction
There are several possible explanations for the observed effects caused by the virioplankton size fraction from environmental samples.
One potential explanation is the presence of cyanophages in the environmental samples. We could demonstrate the presence of plaqueforming units capable to infect M. aeruginosa in 0.2 μm filtrate from algal blooms (Figure 4b + c), and we found no such plaque-forming units in the nonbloom water sample from Lake Bergknappweiher ( Figure 4a). This presence of cyanophages is also reflected in a transcript upregulation when cocultivating M. aeruginosa with the virioplankton size fraction from bloom samples (Figure 3b + c). A further cue suggesting cyanophages as causative agents is the fact that the particle concentrate caused a higher increase in gene transcription than the 0.2 μm filtrate (Figure 3b + c). Viral particles are concentrated in the process of obtaining the particle concentrate unlike other small soluble compounds. Co-occurring cyanophages might cause increased microcystin expression as defense against parasites. As microcystin is mostly located inside the cell (Park et al., 1998;Young, Thomson, Metcalf, Lucocq, & Codd, 2005), a defense function against intracellular pathogens such as bacteriophages is plausible. Microcystin and other secondary metabolites have been shown to reduce the virulence of parasitic fungi to the cyanobacterium Planktothrix (Rohrlack, Christiansen, & Kurmayer, 2013). The observation that autoclaved filtrate increased mcyD transcription in some cases ( Figure 3b

CONFLICT OF INTEREST
None declared.